> Jerry Avins wrote:
>> On 5/3/2010 4:45 PM, glen herrmannsfeldt wrote:
>>> Jerry Avins<jya@ieee.org> wrote:
>>> (snip, someone wrote)
>>>
>
>> I'm puzzled by a familiar circuit. How is that in any way magnetic?
>>
>> Jerry
>
> The new ones aren't magnetic. The ones I worked with in the early 60's
> were. Then around 1968, a fellow named Barrie Gilbert (at Tektronix at
> the time and then Analog Devices) invented an integrated circuit
> version. It beat the heck out of one that I built.
>
> At issue is always the nonlinearities that you don't want. At about the
> same time that Gilbert was publishing his multiplier, I was trying to
> build a more discrete semiconductor version with dual/matched
> transistors. I didn't work where there was a fab and had to buy parts
> off the shelf. I used a matched amplifier approach borrowed from vacuum
> tube designs. You took two matched amplifiers and varied their gain with
> a more or less low frequency signal "X". One amp had a constant input
> and the other had the "Y" input. The one with constant input generated
> an error signal that was applied to both amplifiers so that the gain
> variation with "X" on "Y" would be linearized. I must say it wasn't very
> good but it worked to a point.
>
> Gilbert describes what he did with bipolar junction transistors here:
> http://www.ieee.org/organizations/pubs/newsletters/sscs/jan03/jssc_classic.html
Every so often, the new Thunderbird swallows a post. I'll try again.
I'm familiar with Gilbert cell multipliers, as I wrote to Glen about the
MC1495. We had RCA's CA3101 etc. in the stockroom; very similar. With
feedback, one could divide up to a point, and with an extra op-amp, take
square roots. Notice the similarity to the heart of an OTA*. Linear gain
proportional to a control current is a great tool.
Jerry
______________________
* operational transconductance amplifier
--
"I view the progress of science as ... the slow erosion of the tendency
to dichotomize." --Barbara Smuts, U. Mich.
�����������������������������������������������������������������������
Reply by Fred Marshall●May 4, 20102010-05-04
Jerry Avins wrote:
> On 5/3/2010 4:45 PM, glen herrmannsfeldt wrote:
>> Jerry Avins<jya@ieee.org> wrote:
>> (snip, someone wrote)
>>
> I'm puzzled by a familiar circuit. How is that in any way magnetic?
>
> Jerry
The new ones aren't magnetic. The ones I worked with in the early 60's
were. Then around 1968, a fellow named Barrie Gilbert (at Tektronix at
the time and then Analog Devices) invented an integrated circuit
version. It beat the heck out of one that I built.
At issue is always the nonlinearities that you don't want. At about the
same time that Gilbert was publishing his multiplier, I was trying to
build a more discrete semiconductor version with dual/matched
transistors. I didn't work where there was a fab and had to buy parts
off the shelf. I used a matched amplifier approach borrowed from vacuum
tube designs. You took two matched amplifiers and varied their gain
with a more or less low frequency signal "X". One amp had a constant
input and the other had the "Y" input. The one with constant input
generated an error signal that was applied to both amplifiers so that
the gain variation with "X" on "Y" would be linearized. I must say it
wasn't very good but it worked to a point.
Gilbert describes what he did with bipolar junction transistors here:
http://www.ieee.org/organizations/pubs/newsletters/sscs/jan03/jssc_classic.html
Fred
Reply by Fred Marshall●May 3, 20102010-05-03
Jerry Avins wrote:
> On 5/3/2010 2:07 PM, Tauno Voipio wrote:
>> On 3.5.10 5:49 , Jerry Avins wrote:
>>>
>>>> [The "4 quadrant multipliers" I used were really modulators that
>>>> expected a sinusoidal input for one of the inputs - so in that sense I
>>>> believe they were "tuned". The specs on the output were really tight
>>>> with respect to phase, distortion and gain WRT the "dc" input. They
>>>> were
>>>> electromagnetic devices called "magnetic modulators".)
>>
>> For balanced modulators there were special beam-deflection tubes,
>> a thing like a cross between an amplifier tube and a CRT, e.g.
>> the RCA 7360.
>>
>>> That's new to me. Diode modulators work best when the carrier was strong
>>> enough so that it might as well have been a square wave. The carrier
>>> switches the polarity, washing out any diode drop that the signal might
>> > see.
>>
>> A more modern way of this approcah is to use CMOS analog switches
>> with the selection inputs driven by hard-limited carrier.
>
> I've done that too. my "hard-limited carrier" was the square wave output
> of a CD4046.
>
> Jerry
hmmmmm ... why do I think we've segued into a chopper application to get
around dc amps. Not the same thing.....
Fred
Reply by glen herrmannsfeldt●May 3, 20102010-05-03
Jerry Avins <jya@ieee.org> wrote:
> On 5/3/2010 4:45 PM, glen herrmannsfeldt wrote:
>> Jerry Avins<jya@ieee.org> wrote:
>> (snip, someone wrote)
>>>> [The "4 quadrant multipliers" I used were really modulators that
>>>> expected a sinusoidal input for one of the inputs - so in that sense I
>>>> believe they were "tuned". The specs on the output were really tight
>>>> with respect to phase, distortion and gain WRT the "dc" input. They were
>>>> electromagnetic devices called "magnetic modulators".)
>>> That's new to me. Diode modulators work best when the carrier was strong
>>> enough so that it might as well have been a square wave. The carrier
>>> switches the polarity, washing out any diode drop that the signal might see.
>> See the data sheet for the MC1495. It comes out pretty fast
>> on a google search, from sites like www.alldatasheet.net.
> I'm puzzled by a familiar circuit. How is that in any way magnetic?
Sorry, I wasn't trying to reference the magnetic version.
Putting magnetic modulator into google results in a good number
of hits, mostly using saturating core (and so nonlinear)
transformers.
One reference is to books.google.com for a book on microwave
tube transmitters.
-- glen
Reply by Clay●May 3, 20102010-05-03
On May 3, 5:18�pm, glen herrmannsfeldt <g...@ugcs.caltech.edu> wrote:
> Clay <c...@claysturner.com> wrote:
>
> (snip)
>
> > Actually you are modulating the number of photons emitted by the LED
> > per unit time. Each photon has energy E=h*frequency. So in the case of
> > where your LED is narrowband, then all of the photons have pretty much
> > the same energy. And in the case of a laser diode, then each photon's
> > wavefunctions are also in phase with each other's wavefunctions.
>
> Note that the output of AM and FM radio transmitters is also
> photons, still with E=h*frequency. �So an AM transmitter, as
> with the LED, modulates the number of photons sent per unit time.
>
> Diode lasers have a fairly short (for a laser) coherence time,
> such that the phase can change with time. �(Or consider that
> the bandwidth isn't so narrow as for gas lasers.) �
>
> An individual photon, in the time or frequency EM domain, looks
> like a Gaussian envelope modulation of the appropriate sine.
>
> -- glen
It all comes down to detecting energy although we see the effects of
amplitudes via interference.
You would be suprised at the coherence length of a diode laser these
days. Since they tend to operate in single mode you can make holograms
with a depth 100s of time greater than the diode's cavity size. While
gas lasers due to longer cavities exhibit narrow spectra, the gain
envelope oftens spans more than one mode of the cavity. For example an
HeNe laser will be single mode if its cavity is around 10cm in length,
but such a laser will barely osc. My 10mW HeNe has multiple modes that
caused coherence beats every 27 inches (twice the cavity length).
Clay
Reply by Mark●May 3, 20102010-05-03
On May 3, 4:44�am, Jerry Avins <j...@ieee.org> wrote:
> On 5/3/2010 12:09 AM, glen herrmannsfeldt wrote:
>
> � �...
>
> > So LED output is intensity modulated, not amplitude modulated?
>
> Strictly, yes.
>
> Jerry
> --
> "I view the progress of science as ... the slow erosion of the tendency
> � to dichotomize." --Barbara Smuts, U. Mich.
> �����������������������������������������������������������������������
If I recall correctly, the electrical to light conversion is typically
non-linear and a 1 dB change in the electrical input to the modulator
created a 2 dB change to the light (or was it the other way round)?
And vice versa at the receiver so when you got your electrical signal
back, all was well.
But if you look at it in the light domain, it was NOT the same as if
it were simply a very high frequency RF carrier.
Mark
Reply by Jerry Avins●May 3, 20102010-05-03
On 5/3/2010 4:45 PM, glen herrmannsfeldt wrote:
> Jerry Avins<jya@ieee.org> wrote:
> (snip, someone wrote)
>
>>> Just to shake your cage: the 4 quadrant multiplier used in this mode as
>>> a modulator is a linear system with respect to the modulating signal.
>>> :-) no troll intended really ......
>
>> Sure. Even a plate-modulated class-C amplifier is linear in that respect.
> (snip)
>
>>> [The "4 quadrant multipliers" I used were really modulators that
>>> expected a sinusoidal input for one of the inputs - so in that sense I
>>> believe they were "tuned". The specs on the output were really tight
>>> with respect to phase, distortion and gain WRT the "dc" input. They were
>>> electromagnetic devices called "magnetic modulators".)
>
>> That's new to me. Diode modulators work best when the carrier was strong
>> enough so that it might as well have been a square wave. The carrier
>> switches the polarity, washing out any diode drop that the signal might see.
>
> See the data sheet for the MC1495. It comes out pretty fast
> on a google search, from sites like www.alldatasheet.net.
I'm puzzled by a familiar circuit. How is that in any way magnetic?
Jerry
--
"I view the progress of science as ... the slow erosion of the tendency
to dichotomize." --Barbara Smuts, U. Mich.
�����������������������������������������������������������������������
Reply by glen herrmannsfeldt●May 3, 20102010-05-03
Clay <clay@claysturner.com> wrote:
(snip)
> Actually you are modulating the number of photons emitted by the LED
> per unit time. Each photon has energy E=h*frequency. So in the case of
> where your LED is narrowband, then all of the photons have pretty much
> the same energy. And in the case of a laser diode, then each photon's
> wavefunctions are also in phase with each other's wavefunctions.
Note that the output of AM and FM radio transmitters is also
photons, still with E=h*frequency. So an AM transmitter, as
with the LED, modulates the number of photons sent per unit time.
Diode lasers have a fairly short (for a laser) coherence time,
such that the phase can change with time. (Or consider that
the bandwidth isn't so narrow as for gas lasers.)
An individual photon, in the time or frequency EM domain, looks
like a Gaussian envelope modulation of the appropriate sine.
-- glen
Reply by glen herrmannsfeldt●May 3, 20102010-05-03
Jerry Avins <jya@ieee.org> wrote:
(snip, someone wrote)
>> Just to shake your cage: the 4 quadrant multiplier used in this mode as
>> a modulator is a linear system with respect to the modulating signal.
>> :-) no troll intended really ......
> Sure. Even a plate-modulated class-C amplifier is linear in that respect.
(snip)
>> [The "4 quadrant multipliers" I used were really modulators that
>> expected a sinusoidal input for one of the inputs - so in that sense I
>> believe they were "tuned". The specs on the output were really tight
>> with respect to phase, distortion and gain WRT the "dc" input. They were
>> electromagnetic devices called "magnetic modulators".)
> That's new to me. Diode modulators work best when the carrier was strong
> enough so that it might as well have been a square wave. The carrier
> switches the polarity, washing out any diode drop that the signal might see.
See the data sheet for the MC1495. It comes out pretty fast
on a google search, from sites like www.alldatasheet.net.
-- glen
Reply by glen herrmannsfeldt●May 3, 20102010-05-03
Fred Marshall <fmarshallx@remove_the_xacm.org> wrote:
> Jerry Avins wrote:
(snip, I wrote)
>>> It seems that the usual uses for DSB-SC do use both signs, though.
>>>> Then, if the AM modulation is greater than 100%
>>>> (in some sense yet to be defined) then it depends on how
>>>> it's handled doesn't it?
>>> I asked about this some time ago in this group. If you do
>>> synchronous modulation, or use a four quadrant multiplier as the
>>> mixer, then you get both signs. I don't know in so much detail
>>> how they build modulators or receivers now to know.
>> Synchronous demodulation works, but you need to generate a replica of
>> the carrier to implement it, usually with a PLL. (A narrow band-pass
>> filter around the carrier frequency followed by high gain is the
>> original way.) Once that carrier surrogate is available, it can be
>> directly added to the signal, so a balanced modulator isn't really
>> needed. The choices are then synchronous- or enhanced-carrier
>> demodulation.
> Yep. I was trying to refer to the underlying signals and not the
> mechanics of doing it all that much. But, indeed, a synchronous
> demodulator, or close facsimile thereof, is what would be used for
> DSB-SC.
I was considering the case of normal, not surpressed carrier AM.
Now, you can consider that as DSB-SC modulating a signal with
a DC offset. I haven't visited any AM transmitters lately, and
don't really know how they are currently built.
Otherwise, two commonly (until recently) DSB-SC systems,
the FM stereo subcarrier and the NTSC chroma subcarrier,
separately supply the phase information needed for decoding.
> And, it's the modulators that are difficult of the 4 quadrant
> multiplier variety, not the demodulators. All you need for a
> demodulator is a clocked rectifier sort of circuit - I don't recall what
> they were like but that would be the idea. Well, I guess nothing but a
> switch or two and an inverter. Then a lowpass of suitable bandwidth and
> delay.
Yes that is commonly done in the laboratory style "lock-in amplifier."
The reference signal (carrier) is converted to a square wave and
then to something like a CMOS analog switch.
> Just to shake your cage: the 4 quadrant multiplier used in this mode as
> a modulator is a linear system with respect to the modulating signal.
> :-) no troll intended really ......
(snip)
> glen: I'm not up to speed on current optical signalling methods but
> first thoughts would be:
> - if you really want continuous "analog" modulation then you're stuck
> with having either a bias so that phase is unimportant or you will have
> distortion - or you will have to have a synchronous demodulator. Just
> like any other standard AM situation.
> - most things are digital now so 1 or 0? is usually all we care about -
> and that can be magnitude if amplitude modulation is being used. Then,
> if SNR is good enough multi-level, multi-bit PAM, etc. I don't know
> that +/- has much importance in such a scheme.
Note that it goes back to the Photophone invented by Alexaner
Graham Bell in 1880. (According to Wikipedia, patent 235,199.)
He used a vibrating mirror as the modulator, and crystalline
selenium as the demodulator. I have known of analog modulation
of HeNe lasers, laser diodes, and LEDs, either directly or with
an intermediate carrier.
With the appropriate feedback on a laser, you should be able
to frequency modulate a laser beam. If, for example, you use
a diffraction grating as part of the feedback system and vibrate
the grating with a piezoelectric crystal.
-- glen